trouble running CTFFIND4.1.10 from RELION2.1 on CentOS7.5

Submitted by rbaxter on Fri, 2018-11-02 15:20

Forums

ctffind & ctftilt

Installed RELION2.1 from Github, tried with CTFFIND4.1.10 first the binary then compiled from source (got intel-mkl, wxWidgets etc.) installed in /usr/local/bin

Running through the RELION2.1 tutorial, used unblur/summovie for motion correction. I get stuck at CTF estimation, using all parameters from the manual the job runs but looks like it doesn't find the images. Contents of run.out is:
Using CTFFIND executable in: /usr/local/bin/ctffind
to estimate CTF parameters for the following micrographs:
*
*
...
*
Estimating CTF parameters using Alexis Rohou's and Niko Grigorieff's CTFFIND4.1 ...
1/ 1 sec ..................................,_,"> [oo]
1/ 1 sec ............................................................~~(,_,">
Done! Written out: CtfFind/job005/micrographs_ctf.star

and run.err is:
WARNING: skipping, since cannot get CTF values for
WARNING: skipping, since cannot get CTF values for
...

Yet I can run ctffind from the command line and it seemed to work on the first test mrc file after motion correction.The contents of diagnostic_output.out is:
# Output from CTFFind version 4.1.10, run on 2018-11-02 14:45:43
# Input file: MotionCorr/job002/Micrographs/Falcon_2012_06_12-14_33_35_0.mrc ; Number of micrographs: 1
# Pixel size: 3.540 Angstroms ; acceleration voltage: 300.0 keV ; spherical aberration: 2.00 mm ; amplitude contr
ast: 0.10
# Box size: 512 pixels ; min. res.: 30.0 Angstroms ; max. res.: 7.1 Angstroms ; min. def.: 5000.0 um; max. def. 5
0000.0 um
# Columns: #1 - micrograph number; #2 - defocus 1 [Angstroms]; #3 - defocus 2; #4 - azimuth of astigmatism; #5 -
additional phase shift [radians]; #6 - cross correlation; #7 - spacing (in Angstroms) up to which CTF rings were
fit successfully
1.000000 5405.790527 5305.790527 0.000000 0.000000 -0.172103 1812.479980

And corrected_micrographs.star contains location of motion-corrected mrc files:
MotionCorr/job002/Micrographs/Falcon_2012_06_12-14_33_35_0.mrc
MotionCorr/job002/Micrographs/Falcon_2012_06_12-14_57_34_0.mrc
...

Any advice greatly appreciated.

Alexis Mon, 2018-11-05 14:59

Hello,

The log files you attached look a bit strange - no filenames for the input micrographs are printed out, and the error messages also don't give any names. I guess you didn't setup the job correctly in Relion, but I am no Relion expert so I'm not sure what's going on. The .com file you submitted is also funky - the input filename is not there, just its directory.

The fact that you can run ctffind from the command line without problems also suggests the problem is not with ctffind per se.

Perhaps you'd like to ask relion experts about this problem - they can usually be found on the CCPEM mailing list. Though I suspect you might have missed a step or detail in the tutorial.

I will note that the diagnostic output you attach (which I guess you got from your attempt at running it directly from the command line?) shows that the fit you got is completely wrong - the Thon rings and the fit CTF do not match at all.

Hope this helps,
Alexis